• What are Social Norms? How are They Measured? Gerry Mackie and Francesca Moneti with Elaine Denny and Holly Shakya UNICEF / UCSD Center on Global Justice Project Cooperation Agreement WORKING PAPER 30 September 2014 Gerry Mackie Co-Director, UC San Diego Center on Global Justice Associate Professor of Political Science University of California, San Diego, USA [email protected] http://www.polisci.ucsd.edu/~gmackie/ Francesca Moneti Senior Child Protection Specialist, UNICEF New York, NY USA [email protected] Acknowledgements: We are grateful to many people for inspiration and advice, none of whom are to blame for the document’s shortcomings. We will thank them by name in the final version. 2 Table of Contents I. Introduction 3 Extrasocial Influences and Social Norms 4 What are Social Norms? 6 How are Social Norms Measured? 7 II. What are Social Norms? 8 Interdependent Human Action 13 The Theory of Interdependent Action (Game Theory) 16 The Study of Social Norms 18 Social Approval and Disapproval, and Other Social Influence 23 Typology of Reasons for Behavioral Regularities 25 III. How to Measure Social Norms 38 General Considerations in Measuring Social Norms 42 Identifying Social Norms and their Change in Conversations 48 Simple Inquiries 50 Simple Indicators in DHS or MICS Suggesting the Presence of a Social Norm 52 Lessons from the Reasoned Action Approach 56 An Exemplary Social-Norms, Social Network Study 62 Adapting Individual-Level Stages-of-Change to a Social-Norms Context 64 Community-Level Stages of Change: The Community Readiness Model 67 The Matching-Game Method of Identifying Social Norms 70 Conclusions and Future Research 72 Appendix I: Norms Measurement in Practice 74 Quantitative Methods 74 Mixed-Methods 80 Qualitative Methods 81 Appendix II: Comparing Different Conceptions of Social Norms 87 Works Cited 96 3 I. Introduction The perpetuation of harmful practices, such as caregivers not conversing with infants or female genital mutilation/cutting (FGM/C), and the creation of beneficial new practices, such as exclusive breastfeeding or marriage at an adult age may be due to social motivations. They may involve an entire community’s beliefs and actions rather than simply those of individuals and their families (UNICEF 2010). Social motivations, our focus here, can explain why a behavior – harmful or beneficial – is common in a group. There are many other reasons why population groups might engage in behaviors that are beneficial or harmful to children. Many of the reasons have to do with factors such as the nonsocial environment in which they live or their economic resources. These may determine, for example, their access to health and other services or the availability of clean water. We do not deny the great and often overwhelming importance of these extrasocial influences on behavior of interest. Rather, we focus on the social because it has been somewhat overlooked and misunderstood in development theory and practice. There are three broad categories of beliefs: one’s beliefs about the nonsocial environment (for example, the belief that colostrum is bad for the newborn), one’s beliefs about the social environment, about others’ minds and actions (for example, the belief that my mother-in-law expects me to discard the colostrum), one’s beliefs about one’s own mind (see Adolphs 2009). Beliefs about one’s self includes one’s self-efficacy (Bandura 1997) or, a similar concept, one’s perceived behavioral control over an action (Fishbein and Ajzen 2010). Most development programs give ample consideration to beliefs about the nonsocial environment, and many measure self-efficacy beliefs and their change, but comparatively few consider beliefs about the social environment which are central to the understanding of social norms. Beliefs about what others do, and what others think we should do, within some reference group, maintained by social approval and disapproval, often guide a person’s actions in her social setting. If a harmful practice is social in nature, programs that concentrate on education of the individual, or increase in the availability of alternatives, or provide external incentives, may not be enough. In addition, a program may need to support the clarification, and sometimes the revision, of social expectations of people throughout the entire community of interest. Social-norms theory has not been widely examined in development circles (concepts from Fishbein and Ajzen’s Reasoned Action Approach have been used, but not mostly in health and in the developed world). That’s why the first half of this report attempts a primer on the subject. Social-norms measurement is even less examined. In preparing this essay we examined about 200 different publications and articles on social 4 norms in global development. Most of these studies theorize or detail programming; only about 14% discuss norms-measurements methods. Of these, most discussed qualitative findings or offered baseline and evaluation measures of individual attitudes and behaviors (rather than expectations that members of a group hold of one another, the cement of social norms). Paluck and Ball (2010), reviewing studies of social norms marketing aimed at gender based violence, also find a rarity of specific measurement of social norms change. In the second half of our report we propose some principles for the measurement of social norms and their change, and summarize a variety of ways of doing measurement. Much more work remains to be done by all interested in the topic. Extrasocial Influences and Social Norms This document offers an account of what social norms and other social practices are, with special attention to child well-being, and especially child protection. It also outlines a number of measurement strategies to identify social norms and measure their change over time. Social ways to change social norms are mentioned in passing and by reference (e.g., Cislaghi, Gillespie, and Mackie 2014), and nonsocial ways to change social norms not at all. Call social any influence on action influenced by social norms, and nonsocial any influence on action caused by economic, legal, political, religious, health or educational services, technology change, or other factors. We acknowledge that people are harmed by forces other than social norms, that harmful social norms can have nonsocial origins, and can be ended by nonsocial causes including nonsocial program engagements. For example, suppose that threats to girls’ honor causes families to withdraw them from school, and that this channels the girls to early marriage. Rather than changing the social norm of honor, school buses could be provided by educators, politicians could improve public order, new cellphone technology could reduce personal danger. Any human action in the present is determined by causes at multiple levels – individual, family, community, social, governmental, economic – and present causes are determined by past causes at different levels. For example, unequal gender norms, according to Boserup’s (2007) hypothesis, may have originated or worsened with the shift long ago from hoe to plough agriculture in some regions. The new technology reduced the relative value of female labor, and hence reduced women’s bargaining power in the household. The unequal gender norms persist for centuries after plough societies have moved out of agriculture: although originating in technology change they are maintained as social norms. They are solidified, but may begin to erode under the changed political economy of modernity, as people move to the cities and seek education to participate in an economy that demands education, mobility, and skills that women can perform at least as well as men. That process is reported in an ethnography wonderfully detailing changes in traditional gender norms in a rural Andalucian village from 1963 to 1983 (Collier 1997). Even though the change is pulled by large political and economic forces, humans are not colliding atoms, their agency enacts the change, and movements, policies, and programs can hasten it. 5 The social-ecological model (Bronfenbrenner 2005) directs attention to the multiple levels of causation of an action or practice. Heise (2011), for instance, identifies social norms as one set of causes of intimate partner violence; her evidence-based, ecological model also identifies many more factors at all levels of influence. Other models also open our eyes to multiple causes and multiple ways to respond. The Behaviour Change Wheel (Michie, Atkins, and West 2014) is an example. 6 Michie and coworkers (2013) have also developed a taxonomy of 85 behavior change techniques, organized into 16 clusters by hierarchical analysis! Techniques relating to social norms comprise no more than 10% of the total. The following two subsections preview the report. Material is presented in summary form with almost no citations. Full explication with citations is provided as the report unfolds. What are Social Norms? As a first approximation, a social norm is what people in some group believe to be normal in the group, that is, believed to be a typical action, an appropriate action, or both (Paluck and Ball 2010). A social norm is held in place by the reciprocal expectations of the people within a reference group. Because of the interdependence of expectation and action, social norms can be stiffly resistant to change. The actions of an individual range from the highly independent (like remembering one’s purse on the way out the door in the morning), to the dependent (learning from an acquaintance that a radio show is entertaining,) to the highly interdependent (each driving on the right side of the road because everyone else does). Development thinking has tended to understand individual actions of programmatic interest as being independent; or as being one-way dependent, whereby one person’s action depends on others’, as in the diffusion of innovation. However, there are human 7 actions where what one does depends on what others do, and what others do depends on what one does (many-way interdependence).